<<

SanDiegoLinguisticPapers3(2008) 1131

AspectualandtheAspectPhraseHypothesis* ShinFukuda UniversityofCalifornia,SanDiego Since influential work by Perlmutter (1968, 1970), the standard analysis of English aspectualverbs(EAVs)isthattheyareambiguousbetweencontrolandraisingverbs.In thispaper,Ifirstargueagainstthecontrol/raisinganalysisofEAVsbyshowingthat:(i) thereisnoclearevidencethatanyEAVisthematicand(ii)thereisnoclearevidencethat EAVsformbiclausalsentences.Asanalternative,IproposethatEAVsarefunctional heads projecting their own phrases, or aspect phrases (Travis 1991), in two different positions in a clause: inside the verbal projection (between v and VP) or outside it (immediatelyabove vP).ThedifferenceinthepositionofanEAVisspelledoutastwo differentformsoftheircomplements.WhenanEAVisbetweenvandVP,itisrealized asagerundive.WhenanEAVisabovevP,itisrealizedasan.Iarguethatthe analysisaccountsfortheevidenceusedtomotivatethecontrol/raisinganalysisaswellas previously overlooked differences among EAVs. Further arguments for the proposed analysisofEAVsareprovidedbydatafromaspectualverbsinotherlanguages. 1. Thecontrol/raisinganalysisofEnglishAspectualVerbs 1.1. ArgumentsfortheControl/raisingAnalysis SincePerlmutter(1968,1970),thestandardassumptionhasbeenthatEnglishAspectualVerbs (EAVs)areambiguousbetweencontrolandraisingpredicates.1 TheargumentsfortheraisinganalysiscomefromthefactthatEnglishaspectualverbscan have nonthematic subjects. They allow for expletive subjects (1ab), exhibit active/passive synonymy(2),andpermitidiomchunkstomaintaintheiridiomaticmeanings(3). (1) a. Therebegantobecommotion. (Perlmutter1970:108,(6)) b. Itbegantorain. (Perlmutter1970:109,(7)) (2) a. ThenoisebegantoannoyJoe. b. Joebegantobeannoyedbythenoise. (Perlmutter1970:109,(9)) (3) a. Heedbegantobepaidtourbanproblems. (Perlmutter1970:110,(12)) b. Headwaybegantobemadetowardasolution. (Perlmutter1970:110,(13)) *ThispaperisarevisedandextendedversionofapaperpresentedatWCCFL26atUCBerkeley.Iwouldliketo thankHenryBeecher,IvanoCaponigro,MarkGawron,GeorgeGibbard,AlexdelGiudice,GrantGoodall,PeterJenks, LauraKertz,CynthiaKilpatrick,DanMichel,JohnMoore,MashaPolinsky,HannahRohde,BarbaraStiebels,andthe audienceatWCCFL26fortheirhelpfulcommentsandsuggestions.Needlesstosay,alltheremainingerrorsaremy own.ThisworkwassupportedinpartbyNSFgrantBCS0131946. 1In the terminology used in Perlmutter (1968, 1970), a raising is an intransitive verb that takes a clausal complementandacontrolverbisatransitiveverbwhichrequiresidentitybetweenitsownsubjectandthesubjectof thecomplement,triggeringEqui(valent)NPdeletion. ShinFukuda

Theargumentsforthecontrolanalysiscomefromthefactthatthereseemtobecaseswherethe subjectsofaspectualverbsmustbethematic.First,aspectualverbscanbeembeddedundera subject or object control verb (4a and 4b). Second, aspectual verbs are compatible with the imperative(5). (4) a.Itriedtobegintowork. (Perlmutter1970:111,(20)) b. IforcedTomtobegintowork. (Perlmutter1970:112,(23)) (5) Begintowork. (Perlmutter1970:113,(25)) Inthetransformationalgrammarframework,(4)wastakentoshowthataspectualverbsselectan animate subject in order for ‘EquiNP deletion’ to take place. Likewise, (5) suggests that aspectualverbscanselectasecondpersonsubject(5).Ross(1972)providesfurthersupportfor thecontrol/raisinganalysisofEnglishaspectualverbs,arguingthatonlycontrolverbsallowfor whathecallsAnaphoricComplementDeletion(ACD)(6): (6) a. MaxsuggestedwritingtoSantaClaus,andTeddyagreed/began/approvedtowrite hi m. (Ross1972:576,(8)) b.*It’ssupposedtobemuggytonight,butithasn’tbegunyettobemuggy . (Ross1972:576,(9a)) UsingACDasadiagnostictest,Rossarguesthat ceasecanonlybearaisingverb,unlike stop : (7) Isuggestedthattheynotshriekanymore,sotheystopped/*ceased. (Ross1972:576,(11a)) Rossalsoclaimsthat finish canonlybeacontrolverb,sinceitisincompatiblewiththeweather it (8).Incontrast, stop canbeeitheracontroloraraisingverb,given(7)and(8): (8) Itstopped/*finishedbeingmuggy. (Ross1972:576,(10a)) 1.2. ProblemsfortheControl/raisingAnalysis Although the arguments for the raising analysis of English aspectual verbs remain virtually unchallenged, the arguments for the control analysis are not as decisive and they have been challenged by a number of subsequent studies (Fischer and Marshal 1969, Givón 1973, Newmeyer 1975, Freed 1979, Brinton 1988, Rochette 1999). First, as for the arguments presentedinRoss(1970),ACDin(6)doesnotseemtobeasyntacticprocess,sinceitdoesnot requiresyntacticidentitybetweenthedeletedelementanditsantecedent(i.e.6a).Theclaimthat finish isincompatiblewiththeweather it (8)isnotsupportedbyempiricalevidence,asnaturally occurringexamplesof‘finish’withtheweather it suchas(9)arenotdifficulttofind. (9) a. Thebestpartisthatwhen it ’sfinishedraining,alltheplantsandtreeshavesuddenly…2 b. After it finishedrainingwewentdowntoDiviVillage’snewpool. 3

2http://andrasue.blogspot.com/ 3http://tripreports.visitaruba.com/

12 AspectualVerbsandtheAspectPhraseHypothesis

Moreover,bothNewmeyerandBrintonarguethataspectualverbsaretransparentwithrespectto selectional restrictions, even when they are embedded under a control verb, as in (4). Their argumentisbasedonexamplessuchas(10)(Newmeyer1975:3334,Brinton1988:65). (10) a. Johnaskedhimtolisten/#hear. b. Johnbegantolisten/hear. c. Johnaskedhimtobegin. d.Johnaskedhimtobegintolisten/#hear. (10a)showsthat hear cannotbeembeddedunder ask while listencan.(10b)showsthat begin doesnotimposesuchselectionalrestrictionsand(10c)showsthat begin canalsobeembedded under ask .Assumingthatselectionalrestrictionsarelocal,onewouldexpectthat begintohear should embed under ask , given (10b) and (10c). Instead, (10d) shows that the selectional restrictionconflictbetweenask andhear in(10a)stillhasitseffectsin(10d).Thus,Newmeyer andBrintonarguethataspectualverbsaretransparentwithrespecttoselectionalrestrictionsand cannotbecontrolverbs.Thesecounterargumentsleavetheimperativeevidencein(5)astheonly validargumentforthecontrolanalysisofaspectualverbs.4,5 On the other hand, assuming that aspectual verbs are pure raising verbs raises additional problems. First, this leaves the imperative evidence in (5) unaccounted for. Second, under a raising analysis, sentences with English aspectual verbs must be biclausal, following the standardassumptionthatthecomplementofaraisingverbisaTP.However,evidencediscussed inthe literaturesuggeststhat complements of English aspectual verbs are smaller than a TP. Englishhavebeenanalyzedasconsistingoftheirowntense,basedonthefactthat infinitivescanhaveatimeadverbialthatisinconflictwithanothertimeadverbialmodifyingthe matrixevent,asshownin(11ab)(Bresnan1972,Stowell1982,Pesetsky1991,Bošković1997, Landau2000,Martin2001). (11) a. Yesterday,Johndecidedtoleavetomorrow. b. Today,Johnhopestowinsomeday. Since the infinitival complement in these cases denotes a yettoberealized event, the tense specificationofsuchcomplementshasbeencalledthe‘unrealizedfuturetense’.Aspectualverbs, however,areknowntodisallowthe‘unrealizedfuture’interpretation. (12) a.??Yesterday,Johnbegantoleavetomorrow. b.??Today,thelawceasedtohaveitseffecttomorrow. 4Another argument that Perlmutter presents is the distribution of do so anaphora. Perlmutter claims that a do so anaphorcanreplaceanaspectualverbwhenitisacontrolverb(ia)butcannotwhenitisaraisingverb(ib). (i) a. WarrentriedtobegintoworkandJerrytriedtodosotoo. b.*Oilbegantogushfromthewellandwaterdidsotoo. However,Newmeyerarguesthattheunacceptabilityof(ib)hastodowiththedefinitenessofthearguments,giventhat (ii)isacceptable(Newmeyer1975:31,fn.7). (ii) The oilstoppedgushingfromthewelland the waterdidsotoo. 5Perlmutter(1970)alsopointsoutthataspectualverbstakeanNPcomplement,unliketypicalraisingverbs.However, I focus only on cases with clausal complements in this paper. For recent discussion of aspectual verbs with NP complements,seeThompson(2005)andPylkkänenandMcElree(2006).

13 ShinFukuda

Nonetheless,lackofanindependenttimespecificationintheinfinitivecomplementsofaspectual verbsdoesnotnecessarilymeanthatthesecomplementslacktense.Forinstance,Landau(2000) claimsthatthetenseofinfinitivecomplementsthatcannothavetheirowntimespecificationis anaphoric with the matrix tense. Nevertheless, there is evidence that the complements of aspectualverbslackprojectionsofgrammaticalaspect,whichhasbeenarguedtobelowerthan TP.Akmajian,Steele,andWasow(1979)showthatthecomplementsofaspectualverbscannot encodegrammaticalaspect,neitherprogressivenorperfective(13).Thisisincontrastwithother verbsthatalsotakenonfinitecomplementsbutallowforboth(14). (13) a.*Hebegan[ Progressive beingrunningdowntheroad]. b.*Hebegan[ Perfective tohavefinishedhishomework]. (Akmajianetal.1979:40,(112)) (14) a. We’lltrytomakehim[ Progressive besinging“ComingthroughtheRye”]when…. (Akmajianetal.1979:40,(115a)) b. Iwilltry[ Perfective tohavefinishedthework]bythetime… (Akmajianetal.1979:43,(125)) Giventhehierarchicalorderoftenseandthegrammaticalaspectmarkerssuchas‘be’and‘have’ inEnglish,(13)requiresoneofthefollowingtwo optionstobetrue: (i)the complements of aspectualverbslackthetenseprojectionandgrammatical aspect projectionaltogether; or (ii) theydohavethetenseprojection(anaphoricwiththe matrix tense) but lack the grammatical aspect projection. Due to lack of independent support for (ii), I assume (13) shows that complements of aspectual verbs do not contain a tense projection or grammatical aspect projection. 6Thus,Englishaspectualverbscannotberaisingverbs,giventhestandardassumption thatsentenceswithraisingpredicatesarebiclausal. In sum, the control/raising analysis of English aspectual verbs appears untenable given the evidencethataspectualverbsfailtobehaveascontrolorraisingverbs.AnanalysisofEnglish aspectualverbs,therefore,mustaccountforthefactthattheyarenonthematicwithoutassuming that they are raising verbs. At the same time, it must also account for the fact that English aspectualverbsarecompatiblewiththeimperativewithoutassumingthattheyarecontrolverbs. 2. Proposal:EAVsasHeadsofAspectPhrases InordertoaccountfortheproblematicbehaviorofEnglishaspectualverbs,Iproposethatthey are functional heads which appear in two positions in a clause: between v and VP and immediately above vP. The hypothesis that there is a functional projection within the verbal projection(VPor vP)hasbeenproposedinmanystudies,suchasSportiche(1990,1998),Travis (1991),Koizumi(1994),CollinsandThráinsson(1996)andHallman(2004).GiventhatEAVs encode aspectual information about events, I call their projections aspect phrases, following Travis(1991).Moreover,theprojectionofaspectabove vPiscalled H(igh) Asp (ect)andtheone that is between vandVPis called L(ow) Asp (ect).Finally, I argue that the difference in the positionofEAVsisvisibleinsyntax:thecomplementofHAsp( vP)isrealizedasaninfinitival complement (15a)andthecomplementofLAsp(VP)isrealized as a gerundive complement (15b). 6SeealsoWurmbrand(2006,2007)forargumentsthatEnglishinfinitivesdonothavetenseevenwhentheyhavethe ‘unrealizedfuture’reading.

14 AspectualVerbsandtheAspectPhraseHypothesis

(15) a. Billstartedtorun. b. Billstartedrunning. TP TP V V Bill T’ Bill T’ V V THAspP T vP V V HAsp vP Bill v’ start V V Bill v’ vLAspP 5 V torun LAsp VP start 5 running Theproposedanalysisisdifferentfromthecontrol/raisinganalysisofaspectualverbsinseveral important ways. First, the control/raising analysis assumes a biclausal structure whereas sentences with aspectual verbs are monoclausal in the proposed analysis. Second, the control/raisinganalysisassumesthataspectualverbsassignathetaroletotheirsubjectswhen theyarecontrolverbs.Intheproposedanalysisaspectualverbsarefunctionalheadsandnever bearthetaroles.Third,thecontrol/raisinganalysishasverylittletosayaboutthetwodifferent complementtypes,infinitiveandgerundive.Intheproposedanalysis,thedifferenceinselection oftwocomplementtypesisaconsequenceofthetwopossiblepositionsforaspectualverbsin theproposedanalysis.Crucially,aspectualverbswhichtakeeithertypeofclausalcomplement (e.g. begin , start , continue or cease )canappeareitherasHAsporLAsp,whereasaspectual verbsthattakeonlyagerundivecomplement(e.g. stop and finish )canonlybeLAsp.Inthe reminderofthispaper,ImotivatetheproposedanalysisforEAVswithbothlanguagespecific andcrosslinguisticarguments. 3. ArgumentsfortheAspectPhraseAnalysis 3.1. TheSizeofInfinitivesandGerundives ImmediateconsequencesofanalyzinginfinitivesandgerundivesunderEAVsas vPsandVPsis thatsuchananalysisaccountsforboth(i)thelackofevidenceforatenseprojection(12)and(ii) the evidence for the lack of the grammatical aspect projection (13) in the complements of aspectualverbs.Ifinfinitivesandgerundivesunderaspectualverbsare vPsandVPs,respectively, neithertensenorgrammaticalaspectcanbepresentinthesecomplements. There is also evidence that infinitives under aspectual verbs are structurally larger than gerundives in the same environment. The evidence comes from the interpretation of certain .Englishhasadverbssuchas stupidly thatcanbeambiguousbetweenaspeakeroriented reading,whichisgenerallyassociatedwitharelatively high syntactic position, anda manner reading, which is generally associated with a relatively low syntactic position. Interestingly, whensuchanambiguousoccurswithintheclausalcomplementofanaspectualverb,its possibleinterpretationsdifferaccordingtothenatureofthecomplement.Fournativespeakers that I consulted interpreted stupidly aseitheraspeakerorientedadverbonlyorasambiguous betweenthetwopossiblereadingswhenitimmediatelyprecededaninfinitive(16a),whereasthe

15 ShinFukuda

sameadverbwasonlyinterpretedasamanneradverbwhenitimmediatelyprecededagerundive (16b). (16) a. …foundeveryonearoundmegrewquietasIbegan stupidly tosay whatIreally think.7 b. …foundeveryonearoundmegrewquietasIbegan stupidly saying whatIreally think. First, the observation that the speakeroriented reading of stupidly is only available with an infinitive supports the claim that infinitives are structurally larger than gerundives. Speaker orientedadverbsaregenerallyassociatedwiththeCPdomain(Alexiadou1997,Cinque1999). However,theevidencepresentedearlier((12)and(13))stronglysuggeststhatinfinitivesunder EAVsareunlikelytobeevenTPs,muchlessCPs.Giventhegrammaticalityof(16a),Iargue that an infinitive under EAVs is a vP, based on assumptions that speakeroriented adverbs minimallyrequireapropositionastheircomplementanda vPisasmallestsyntacticunitwith which a complete proposition can be syntactically represented (see Bale 2007 for a relevant discussion).Second,theobservationthatthesameadverb stupidly canonlybeinterpretedasa manner adverb with a gerundive (16b) suggests that a gerundive cannot be a vP under the adoptedassumptions.AreasonableanalysisappearstobethatagerundiveunderEAVsisaVP, given that manner adverbs are generally associated with VPs. Finally, the availability of the mannerreadingof stupidly withtheinfinitiveisconsistentwiththeanalysisthataninfinitiveisa vP,sincea vPembedsaVP. FurthersupportfortheclaimthatinfinitivesunderEAVsarelargerthangerundivesinthe sameenvironmentcomesfromdifferencesintheirselectionalrestrictions(Bolinger1968,Freed 1979, and Brinton 1988). The examples in (17) below show that gerundives under aspectual verbs force a single event interpretation of the embedded verb, making these sentences very awkward.Incontrast,infinitivesinthesameenvironmentdonotimposesuchaninterpretation andallowforareadinginwhichthesameeventisrepeatedmultipletimes. (17) a. Thatneverceasestoamaze/??amazingme. b. Thatstudentcontinuedtofallasleep/??fallingasleepinmyclass. Asharpercontrastcanbeseenwithembeddedstatives.Gerundivesunderaspectualverbssimply cannothaveastativeverb,unlikeinfinitives. (18) a. Theproblemceasedtoexist/*existing. b. Norabegantoknow/*knowingrightfromwrong. Thus,theevidencesuggeststhatinfinitivesunderaspectualverbsarelargerthangerundivesin thesameenvironment.Specifically,thepossibleinterpretations of the adverb stupidly suggest thatinfinitivesare vPs,whilegerundivesareVPs. 3.2. QuantifierScope Thesecondargumentfortheproposedanalysisistheambiguityresultingfromtheinteraction betweenaquantifierinsubjectpositionandaspectualverbs.May(1985)pointsoutthatraising predicatesinteractwithaquantifierinsubjectpositionandcreateambiguity.In(19)below,a 7http://www.batgirl.com/archives/000528.php

16 AspectualVerbsandtheAspectPhraseHypothesis

quantifier someone is ambiguous between a specific reading (19a) and an existential reading (19b). (19) SomeonefromNYislikelytowinthelottery. a. ThereisapersonfromNYwhoislikelytowinthelottery.(specific) b. ItislikelythatapersonfromNYwillwinthelottery.(existential) Undertheraisinganalysis,thequantifierinsubjectposition, someone ,isbasegeneratedasthe embedded subject and undergoes A movement to become the matrix subject. The existential interpretationof someone obtainsbecause someone isunderthescopeof belikely initsbase generatedposition,andthespecificinterpretationof someone obtainsbecause someone endsup takingscopeover‘belikely’aftermovingtothematrixsubjectposition. Theproposedanalysispredictsthatasimilarambiguityshouldobtainwithanaspectualverbin HAsp,sincethesubjectisunderthescopeofHAspinitsbasegeneratedposition(Specof vP), butitendsuptakingscopeoverHAsponceitmovestoSpecofTP.Ontheotherhand,suchan ambiguity is not expected with LAsp, which is lower than vP. This prediction is borne out. Accordingtofournativespeakersconsulted,‘someone’isambiguousbetweentheexistentialand specificreadingwithaninfinitivecomplementunderanaspectualverbinHAsp(20a),whereas noambiguityisfoundwithagerundivecomplementunderanaspectualverbinLAsp(20b). 8 (20) a. SomeonefromNYstartedtowinthelottery.(infinitive=HAsp) i) someone>start(specific) ii) start>someone(existential) b. SomeonefromNYstartedwinningthelottery.(gerundive=LAsp) i) someone>start(specific) ii)??start>someone(existential) Therefore,thecontrastinthepossiblescopeinteractionsbetween(20a)and(20b)supportsthe proposedanalysisofEAVs. 3.3 QuantifierFloat Cable(2004)arguesthatcertainEnglishpredicates,suchas try ,instantiaterestructuringwhen they take a gerundive complement. One of his arguments for the restructuring analysis of predicatessuchas try ,asopposedtononrestructuringpredicatessuchas prefer ,isdistributionof floatingquantifiers.UnderthebareVPcomplementanalysisofrestructuring(SeeWurmbrand 2001 for references and a comprehensive discussion of the VP complement analysis of restructuring),agerundivecomplementofarestructuringissubjectless.Assumingthat afloatingquantifierall mustbeinalocalandccommandingrelationwiththeNPwithwhichit isassociated, all ispredictedtobenotlicensedwithabareVPcomplementunderarestructuring predicate. 9 Cable shows that the prediction is borne out, based on the contrast in the 8Thescenariogiventothenativespeakersfor(20)isthefollowing.ThestateofNYestablishedalotterytenyearsago. Inthefirstfewyears,peoplefromotherstateskeptwinningthelottery.Fiveyearsago,someonefromNYfinallywon thelottery,andsincethen,thelotteryhasalwaysbeenwonbysomeonefromNY. 9Thispredictionisbasedonthedistributionof‘all’ anditholdswhetheroneassumesthestrandinganalysisorthe adverbialanalysisoffloatingquantifiers.SeeBobaljik(1998,2003)andFitzpatrick(2006)forrecentdiscussionsof thedifferentapproachestofloatingquantifiers.

17 ShinFukuda

grammaticalityof all immediatelyprecedingthegerundivecomplementofprefer (21a)and try (21b): (21) a. Wewouldprefer all ridingthetraintogether. b.*Wetried all ridingthetraintogether. The proposed aspect phrase analysis of EAVs also predicts that a floating all would be ungrammatical with a gerundive complement of EAVs, which is a subjectless VP, while a floating all wouldbegrammaticalwithaninfinitivecomplementofEAVs,whichisa vP.This predictionisborneoutwiththeresultsofonlinesearches for naturally occurring examples of floating all inthecomplementofEAVs.Usingthesearchengine Google, naturally occurring examplesof all adjoinedtoaninfinitiveorgerundivecomplementofEAVsweresearchedfor.In ordertofindrelevantexamples,twostrings, aspectualverb+to+all and aspectualverb+all+verb , withanEAVineitherpresenttenseorpasttenseweresearchedfor,usingthreeEAVsthattake eitherinfinitiveorgerundivecomplement: begin , continue and start .Thefirsttwohundredhits ofeachofthefourstringsforthreeEAVs(atotalof2400hits)weremanuallyexaminedtolook fortherelevantexamples.Table1belowsummarizestheresults: Table1: ResultsofOnlineSearchfor‘floating all ’embeddedunderaspectualverbs present+to+all+ past+to+all+ present+all+ past+all+ verb verb verbing verbing (token/type) (token/type) (token/type) (token/type) begin 100/44 83/43 0/0 1/1 continue 21/13 4/4 0/0 0/0 start 19/12 101/44 0/0 0/0 total 140/69 188/91 0/0 1/1 The results show a clear difference between infinitive and gerundive with respect to their compatibility with all . While naturally occurring examples of all adjoining to an infinitive complement under an aspectual verb are abundant (total 328), there is only one naturally occurringexampleof all adjoinedtoagerundivecomplementunderanaspectualverb.10 (22a)to (22d)aresomeoftheexamplesof‘all’withaninfinitivecomplement.(23)istheonlyexample of all withagerundivethatwasfound: (22) a. ..andintheendgetintoahugefightwheretheybeginto all killeachother.11 b. wecancontinueto all worktogethertorecognisethechallengeandbuildon..... 12 c. Thepossibleknockagainstthisalbumisthatthesongsstartto all soundthesame. 13 d. Thestoriesstandontheirown,sharingsomecharacters,butasthemovieprogresses theystartto all knittogetherforthebigfinale. 14 10 Although the contrast is clear in the results of the online search, native speakers’ judgments suggest that the differenceissubtle.AmongthefournativespeakerstowhichIaskedtheirjudgmentsonsentenceswith‘all’adjoined toaninfinitiveandagerundiveunderanaspectualverb,onespeakerfoundnosignificantdifference,anotherspeaker mentioned that the gerundive examples are only slightly more unusual, while two other speakers reported the gerundiveexamplesareclearlylessacceptable. 11 http://en.wikipedia.org/ 12 http://www.naturebase.net/ 13 http://www.amazon.com/ 14 http://www.cinegeek.com/

18 AspectualVerbsandtheAspectPhraseHypothesis

(23) …aroundthesametimeaswhenwebluekidsbegan all sittingonthesamesideofthe roomsowecouldallbetogetherforourmorning ... 15 Thus,thedistributionoffloating allunderEAVsprovidesanothersupportingargumentforthe proposedanalysisofEAVs. 3.4. LongPassive FurthersupportfortheproposedanalysisofEnglishaspectualverbsisprovidedfromphenomena commonlycalledlongpassive.Longpassiveispassivization of an embedded object with the passivemorphemeappearingonlyonthematrixpredicate.Longpassivehasbeenobservedwith aspectualverbsandotherrestructuringverbsinlanguagesasdifferentasSpanish(Aissenand Perlmutter 1976, 1983), German (Wurmbrand 2001), Japanese (Shibatani 1973, Nishigauchi 1993, Matsumoto 1996), Chamorro (Chung 2004) and Kannada (Agbayani and Shekar, to appear). InJapanese,aspectualverbscanbeclassifiedintothreegroupsbasedontheirbehaviorwith respecttopassive.Theaspectualverb owar ‘finish 1’onlyallowsanembeddedpassive(24a); anotheraspectualverb, oe ‘finish 2’,onlyallowslongpassive(24b),whiletwootheraspectual verbs, hajime ‘begin’ and tsuzuke ‘continue’, allow both options (25) (Shibatani 1973, Nishigauchi1993,Matsumoto1996): 16 (24) a. Rombunga[ti kak are ] owar (*are) ta paperNOM [ti write PASS ] finish 1(*PASS ) PERF ‘Thatpaperfinishedbeingwritten.’(embeddedpassiveonly) b. Rombun iga[t i kaki (*are) oe ] rare ta paper iNOM [t i write (*PASS ) finish 2]PASS PERF ‘Thatpaperfinishedbeingwritten.’(longpassiveonly) (25) a. Rombunga[ti kak are ] hajime/tsuzuke ta paperNOM [ti write PASS ] begin/continue PERF ‘Thatpaperbegan/continuedtobewritten.’(embeddedpassive) b. Rombun iga[t i kaki hajime/tsuzuke ] rare ta paper iNOM [t i write begin/continue] PASS PERF ‘Thatpaperbegan/continuedtobewritten.’(longpassive) While previous analyses attempted to account for the pattern in (24) and (25) based on the assumption that Japanese aspectual verbs are either control or raising verbs that may involve restructuring or reduced complements (Nishigauchi 1993, Kageyama 1993, 1999, Matsumoto 1996,Koizumi1998),IproposedinFukuda(2006,2007)thattheseaspectualverbsareheadsof functionalprojectionswhichcanappearintwodifferentpositions:LAsp(below vP)andHAsp (above vP).Undersuchananalysis,theregularembeddedpassive is the only option with an aspectualverbinHAsp,sincethepassivemorphememustprecedetheaspectualverb,assuming thatthepassivemorphemeoccupiesthe vposition(Krazter1994,1996,Chomsky1995among

15 http://uncpress.unc.edu/ 16 Abbreviations: NOM =nominative, ACC=accusative, PASS =passive, PERF =perfective.

19 ShinFukuda

others)(26a).Incontrast,longpassiveistheonlyoptionwithanaspectualverbinLAsp,since thepassivemorphemeisabovethepositionoftheaspectualverbandmustfollowit(26b). (26) a. HAsp=onlyembeddedpassive b. LAsp=onlylongpassive TP TP V V

NP iT’NP iT’ V V HAspPTvPT V V vP HAsp LAspPPassive V V VPPassiveVP LAsp V V

ti V V+PASS +ASP tiV V+ASP +PASS BothWurmbrand(2001)andCinque(2003)independentlyproposedsimilaranalysesforlong passiveinGermanandinRomancelanguages,respectively. 17 IftheproposedaspectphraseanalysisofEAVsisontherighttrack,thecrosslinguisticdata justdiscussedsuggestthatthelongpassiveshouldalsobegrammaticalwithEAVsinLAsp.In fact,onedoesnothavetotrytoohardtofindexamplesofapparentlongpassivesinEnglish: (27) a. When thepiesandcakeswerefinishedbaking,itwasabout…18 b. Defendantwaiteduntilthesheetswerefinishedwashingand…19 c. TheRV9AwasfinishedpaintinglastSunday.20 ItappearsthattherearetworestrictionstotheapparentcasesoflongpassiveinEnglish.First, longpassiveisgrammaticalonlywith finish ,andnotwithotherLAspverbs(28). (28) Thesecakeswerefinished/*werecontinued/*werebegan/*werestoppedbaking. Second, not alltransitiveverbs form a grammatical long passive sentence under finish . (29a) showsthatlongpassiveisungrammaticalwithcertainactivitypredicates,suchas watch ,whileit isgrammaticalwithanotheractivitypredicate, bake(29b): (29) a.??Thesemovieswerefinishedwatching. b. Thesecakeswerefinishedbaking. Itturnedoutthatsimilarobservationshavebeenmadewithlongpassivewithaspectualverbs in other languages. The first restriction, that long passive is restricted to finish , finds its 17 AnimportantdifferencebetweenWurmbrand(2001)andCinque(2003)isthatall‘restructuringverbs’(which includeaspectualverbsthatallowforlongpassive)arefunctionalheadsinCinque,whereasWurmbrandmakesa distinctionbetweenfunctionalandlexicalrestructuringverbs. 18 http://www.allairevillage.org 19 http://www.sconet.state.oh.us 20 http://www.avsim.com

20 AspectualVerbsandtheAspectPhraseHypothesis

equivalentinlongpassiveinSpanish.AissenandPerlmutter(1983)notethatlongpassiveis grammaticalinSpanishonlywithcompletiveaspectverbs,suchas terminar ‘finish’and acabar de ‘finish’: (30) a. Estas paredes están siendo terminadasde pintar(porlosobreros) these walls are being finishedtopaint (bytheworkers) ‘Thesewallswerefinishedbeingpainted(bytheworkers),’ b. Lascasas fueronacabadas de pintar (porlosobreros) thehouseswere finished to paint (bytheworkers) ‘Thehouseswerefinishedbeingpainted(bytheworkers).’ (AissenandPerlmutter1983;390391,(P33b)and(P34b)) OtherrestructuringpredicatesinSpanish,whichotherwiseshowmonoclausalbehavior,donot allowlongpassive.Moreover,inJapanese,oneofthecompletiveaspectverbs, oe ‘finish 2’isthe onlyoneamongthefouraspectualverbsdiscussedearlierthatisrestrictedwithlongpassiveas seenin(24b).Theseobservationssuggestthatsomethingaboutaspectualverbswithcompletive aspectmakesthemparticularlysuitableforlongpassive.Thus,thefactthatonly‘finish’allows theapparentlongpassiveinEnglishisprobablynotacoincidence. ThesecondobservationsuggeststhatEnglishlongpassiverequirescertainsemantictypesof objects.Whilebothwatch and bake areactivitypredicates,onewayinwhichtheseverbsdifferis that only an object of ‘bake’ undergoes a change of state. Whereas cakes come into their existenceinbakingevents,moviesdonotchangetheirstatusinwatchingevents.Infact,the grammaticalexamples oflong passive that I found all involve transitive verbs whose objects undergoachangeofstate,suchassheetsthatwerewashed(27b)andanairplanebeingpainted (27c). Whydoeslongpassiverequireanobjectthatundergoesachangeofstate?Isuggestthatthe relevantfeatureofthepassivizedobjectinlongpassiveisitsabilitytodelimiteventsandlong passive is grammatical only with objects that delimit events. Objects of transitive verbs that undergoachangeofstatedelimiteventsembeddedunderanaspectualverb.Thus,theymake grammaticallongpassivesentences.Ontheotherhand,objectsoftransitiveverbsthatdonot undergoachangeofstatedonotdelimiteventsembeddedunderanaspectualverb.Therefore, theyfailtomakegrammaticallongpassive. DatafromItaliandiscussedinCinque(2003)appeartosupportthishypothesis.Cinqueshows that, in Italian, long passive is grammatical with inceptive aspectual verbs (i.e. ‘begin’) and completiveaspectverbs(i.e.‘finish’)butonlymarginally acceptable with continuative aspect verbs(i.e.‘continue’).Withthoseaspectualverbsthatarecompatiblewithlongpassive,Cinque showsthatthereisanadditionalconstraintongrammaticallongpassive:thepassivizedobject mustbequantized.Withthesameaspectualverbs,longpassiveisgrammaticalwithaquantized object due case ‘twohouses’(32a)butungrammaticalwithabarepluralobject case ‘houses’ (32b): (32) a. Furuno iniziate/?cominciate a costruire solo due case were begun 1/begun 2 to build only two houses ‘Onlytwohouseswerebeguntobebuilt.’ b. *furuno iniziate/comninciate a construire case were begun 1/begun 2 to build houses ‘Houseswerebeguntobebuilt,’ (Cinque2003;56,(10))

21 ShinFukuda

Withaquantizedobject, duecase ‘twohouses’,thebuildingeventin(32a)hasaclearendpoint. Inotherwords,theeventembeddedundertheaspectualverbisdelimited.Ontheotherhand,the barepluralobjectcase ‘houses’doesnotdelimittheembeddedeventin(32b).Thus,itappears that, in both English and Italian, the passivized object in long passive must be capable of delimitinganeventembeddedunderapassvizedaspectualverb. Therefore,thetworestrictionsobservedwiththeapparentlongpassiveinEnglish,(i)thatonly finish allowsforlongpassiveand(ii)thatonlycertaintransitiveverbsformgrammaticallong passive sentences, turned out to show that English long passive shares similarities with long passive in other languages such asJapanese, Spanish, and Italian.While a careful andfully developedanalysisofEnglishlongpassiveisclearlynecessary,theinitialobservationsaboutthe apparentlongpassiveinEnglishprovideadditionalsupportfortheproposedanalysisofEAVs, accordingtowhichEAVsappearintwodifferentpositionsinaclause,oneofwhichisbelow v. 4. ImperativeRevisited Sofar,Ihavepresentedseveralargumentsfortheaspect phrase analysis of Englishaspectual verbs. There is an important question that has been left unanswered: if, as I have argued, aspectual verbs are functional heads with no theta role, how are they compatible with the imperative? Isuggestthatanexplanationcanbefoundinthetwopositionsavailabletoaspectualverbs. WithLAsp,anaspectualverbisfirstcombinedwithVP,forminganaspectphrase(33b).This aspectualphraseisthencombinedwith v(33c),whichintroducesthesubjectasitsspecifier(33d). (33) a. VPb.LAspP c.v’ d. vP 4  V  V  V run LAsp VP v LAspP Bill v’ start 5 6 6 running startrunningstartrunning WithHAsp,ontheotherhand,av’isfirstformedwith vanditscomplement,VP(34b),andthe subjectisthenmergedtothisstructure, v’(34c),beforeanaspectualverbisintroduced(34d). (34) a.VPb. v’ c. vPd.HAspP 4  V  V  V run vVP Bill v’ HAsp vP 5 5 start V torun torunBill v’ 5 torun FollowingKratzer(1994,1996),Iassumethat v(Voice inKratzer)createsapredicationrelation betweentheexternalargumentinSpecof vPanditscomplement.Thus,onlywithLAspisthe subjectpredicatedoftheaspectphrase.Inotherwords,‘Bill’isthesubjectof‘startedrunning’in (33),butin(34)itisthesubjectonlyof‘torun’.Thus,onlyLAspin(33)hasthe‘thematic’ interpretation,whichinspiredtheoriginalcontrolanalysisofaspectualverbs. Now, the imperative requires an appropriate external argument (e.g. agent). Given the assumptionthat vPiswheretheexternalargumentisintroduced,Iproposethattheimperative formation targets vP. 21 Under such an analysis, the current proposal predicts that only an aspectualverbinLAspwithagerundivecomplementshouldbecompatiblewiththeimperative, since LAsp is part of vP.AnaspectualverbinHAspwithaninfinitivalcomplement is not 21 Unlessnegationisinvolved(seePotsdam,inpress,forarecentanalysisofEnglishnegativeimperatives).

22 AspectualVerbsandtheAspectPhraseHypothesis

expectedtobefelicitouswiththeimperative,sinceHAspisabove vP.Elicitationswithfifteen nativespeakersshowthatthisisindeedthecase.Theywerefirstgivenasituationwhichcanbe followedbyanimperativesentence,suchas(35),andaskedtochoosebetweenanimperative sentencewithagerundive(35a)andwithaninfinitive(35b).93%orfourteenoutofthefifteen nativespeakerschosetheexamplewithagerundiveasthemorenaturaloption. (35) Situation:aftergivingdirectionstoagroupofstudentswhoareabouttowriteaninclass essay,theproctorsays: a. Beginwriting!!(93.3%or14/15) b. Begintowrite!!(6.7%or1/15) Therefore, the proposed analysis makes the right prediction about imperative formation with EAVs. In contrast, the traditional control/raising analysisofEAVsdoesnotseemtoofferan accountforthecontrastwithtwodifferentformsofthecomplementofEAVswithrespectto imperative. 5. FurtherArgumentsfromOtherLanguages Inthissection,Ipresentfurtherargumentsforthe aspect phrase analysis of EAVs from data concerning aspectual verbs in two other languages: German and Basque. German data are discussedinWurmbrand(2001),inwhichthecontrolandraisinginterpretationsofanaspectual verb beginnen ‘begin’arearguedtocorrespondtotwodifferent syntactic positions, based on distribution of neighboring elements, such as modals. Basque data come from Arregi and MolinaAzaola(2004),inwhichtwoaspectualverbs, hasi ‘begin’and amaitu ‘finish’,areargued tooccupytwodifferentsyntacticpositions,basedonlongdistanceagreementphenomena. 5.1. German(Wurmbrand2001) Wurmbrand(2001)arguesthatcontrolandraisingverbsinGermanoccupydifferentpositionsin a clause. While German raising verbs occupy the position where auxiliary verbs are found, GermancontrolverbsarefoundastheheadofVP(Wurmbrand2001:206). (36) AuxP V Aux’ V ModPAux raisingverbs V Mod’ V vPMod V v’ V VPv V V controlverbs

23 ShinFukuda

Underthisanalysis,aspectualverbsthathavebeenarguedtobeambiguousbetweencontroland raisingareanalyzedtoappearineitherofthesetwopositions(eitherastheheadofAuxPorVP), whileunambiguouscontrolandraisingverbsarerestrictedtooccurastheheadofVPandthe headofAuxP,respectively. Inordertoshowdifferencesinthestructuralpositionofcontrolandraisingverbs,Wurmbrand contraststhedistributionofunambiguousraisingverbs,suchas sheinen ‘seem’and pflegen ‘be useto’,andambiguousverbs,suchas versprechen ‘promise’, drohen ‘threaten’andanaspectual verb, beginnen ‘begin’.First,Wurmbrandshowsthattheunambiguousraisingverbscannotbe embedded under amodal,althoughtheycanembedamodal. 22 (37) a.*Morgan wird /dürfte er di Stadt zu verlassen sheinen Tomorrowwill/might he the town to leave seem ‘Hemightseemtobeleavingthetowntomorrow.’ (Wurmbrand2001:207,(168a,168c)) b. Sieschien zu Hause arbeiten zu müssen/können sheseemedat home work to must/can ‘Sheseemedtohaveto/beabletoworkathome.’ (Wurmbrand2001:207,(169d,169e)) Incontrast,ambiguousverbscanbeembeddedunderamodalverb.However,whentheyare, theycanonlybeinterpretedascontrolverbs(i.e.theepistemicinterpretationof‘promise’and ‘threaten’arenotavailable). (38) Er muβ ein guter Vater zu warden versprechen/drohen He must a good father to become promise/threaten ‘Hemustpromise/threatentobecomeagoodfather.’ (Wurmbrand2001:209,(172c,172d)) Second, both the unambiguous raising verbs and the ambiguous verbs allow passive in the embedded clause ((39a) and (39b)). However, once an ambiguous verb is embedded under a modal,embeddedpassivebecomesungrammatical(39c). (39) a.?DieStadt began zerstört zu werden thetown began destroyed to AUX Pass ‘Thetownbegantogetdestroyed.’(Wurmbrand2001:211,(174c)) b. DerKaviarschien gegessen worden zu sien thecaviar seemedeaten been to be ‘Thecaviarseemedtohavebeeneaten.’(Wurmbrand2001:208,(170c)) c.*Die Stadt muss/kann zerstört zu werden beginnen thetown must/can destroyed to AUX Pass begin ‘Thetownmust/can/maybegintogetdestroyed.’ (Wurmbrand2001:211,(176a,176b)))

22 AccordingtoWurmbrand,onlydeonticmodalscanbeembeddedunder sheinen ‘seem’,which,underheranalysis, occupiesthepositionwhereauxiliaryverbsandepistemicmodalsoccupy.

24 AspectualVerbsandtheAspectPhraseHypothesis

Third,whiletheunambiguousraisingverbsdonotpassivize(40a),theambiguousaspectualverb beginnen ‘begin’canpassivize,whichcanbeimpersonalpassive(40b)orlongpassive (40c). (40) a.*DerKaviar wurde zu essen geschient/geschienen thecaviar was to eat seemPART a/seemPART b ‘Itseemedthatsomebodyatethecaviar.’(Wurmbrand2001:208,(170a,170b)) b. Eswurde begonnen den Wagen zu reparieren it was beganPART the carACC to repair ‘Theybegantorepairthecar.’( impersonalpassive )(Wurmbrand2001:212,(178b)) c. Der Wagen wurde zu reparieren begonnen the car was to repair begunPART ‘Theybegantorepairthecar.’( longpassive )(Wurmbrand2001:213,(180a)) Here, beginnen ‘begin’ differs from two other ambiguous verbs that Wurmbrand discusses, versprechen ‘promise’and drohen ‘threaten’,asonly beginnen ‘begin’allowsforlongpassive. Withthetwootherambiguousverbs,onlyimpersonalpassiveisgrammatical(Wurmbrand2001, 213). Finally, an impersonal passive sentence with an ambiguous verb, as in (40b), can be embeddedunderamodal(41).However,inthisparticularenvironment,theambiguousverbcan onlybeinterpretedasacontrolverb,aswasthecasewiththe‘disambiguated’instancesofthe ambiguousverbs,seenin(38)above. (41) Es muss sofort begonnen wardenden Wagen zu It must immediately begin Aux Pass the carACC to reparieren repair ‘Theymustbeginimmediatelytorepairthecar.’(Wurmbrand2001:213,(179b)) Wurmbrandarguesthatalloftheaboveobservationsshowthatcontrolandraisingverbsoccupy two different structural positions, as illustrated in (36). The unambiguous raising verbs can embedamodalbutcannotbeembeddedunderamodal,asin(37),becauseraisingverbsoccupy apositionashighasorhigherthanwheremodalsappear.Whenanambiguousverbisembedded underamodal,asin(38),theverbcanonlybeinthelowerposition,whereitisinterpretedasa control verb. Moreover, both the unambiguous raising verbs and the ambiguous verbs are expectedtohaveembeddedapassivecomplement((39a)and(39b)),sinceraisingverbsoccupy the position higher than vP,wherethepassivemorphemeisassumedtooccupy.Yetoncean ambiguousverbisembeddedunderamodal,asin(39c),itcanonlybeinthelowerposition, whereitisinterpretedasacontrolverb.Whenanambiguousverbisinthelowerposition,there isno vPprojectionbelowit.Thus,itfollowsthatanambiguousverbcannotembedapassive complementin(39c).Moreover,whiletheunambiguousraisingverbsareabove vPandtherefore theydonotpassivize(40a),theambiguousverbsarebelow vPandthereforetheyareexpectedto passivize ((40b) and (40c)). Finally, since only an ambiguous verb in the lower position is expectedtopassivize,the‘passivized’aspectualverbisexpectedtoembedunderamodal(41). Ontheotherhand,ifweweretomaintaintheassumptionthatcontrolandraisingverbsappear inthesameposition,i.e.theheadofVP,theabove observations wouldbe problematic. One wouldhavetoassume,forinstance,thatthesedifferencesderivefromdifferencesineachverb’s compatibility with modals and passives. However, such an approach would have to be quite

25 ShinFukuda

complex,since,undersuchascenario,theambiguousverbs’compatibilitywithpassivewould havetochange,dependingonthepresence/absenceofamodal. Thus,Wurmbrandconvincinglyshowsthatthesamepredicatesareinterpretedascontroland raising verbs depending on their syntactic positions in German. As such, the German data discussed by Wurmbrand support the proposed analysis of EAVs, according to which the control/raising ambiguity with EAVs is argued to be due to two positions in a clause where EAVsmayoccur. Moreover,Wurmband’sdataalsosuggestthat beginnen ‘begin’mayoccupyapositionthatis lower than one that two ambiguous verbs, versprechen ‘promise’ and drohen ‘threaten’ may occupy.Recallthatbeginnen ‘begin’istheonlyverbswhichallowsforbothimpersonalpassive andlongpassive.Thetwootherambiguousverbs, versprechen ‘promise’and drohen ‘threaten’, onlyallowforimpersonalpassive.Inimpersonalpassive,theembeddedobjectislicensedwith accusativecasedespitethefactthatthematrixverbispassivized.Undertheassumptionthatitis vthatprovidesaccusativecase,thismeansthattheambiguousverbsembeda vPinimpersonal passive,asin(42). (42) Eswurde begonnen [vP den Wagen zu reparieren] it was beganPART [vP the carACC to repair] Onewaytoaccountforthestructurein(42)isto analyze impersonal passive as involving a clausalcomplementation(i.e.abiclausalstructure)inwhich versprechen ‘promise’and drohen ‘threaten’occupythepositionforlexicalverbs,V,andtakea vPcomplement.Ontheotherhand, inlongpassive,theembeddedobjectmovestothematrixsubjectposition,presumablybecause accusativecaseisnotavailableinalongpassive sentence. Underthe same assumption about accusativecasementionedabove,thismeansthatthecomplementof beginnen ‘begin’inlong passivemustnotcontain v,i.e.itisaVP. (43) Der Wagen wurde [VP zu reparieren] begonnen the car was [VP to repair] begunPART Thereareatleasttwo possible analysesoflong passive in (43). One is to assume that long passiveresultswhen beginnen ‘begin’takesaVPcomplement.Undersuchananalysis, beginnen ‘begin’occupiesVjustlikeitisassumedtodoin(42)andthestructureisstillbiclausal.The difference between beginnen ‘begin’ on one hand and versprechen ‘promise’ and drohen ‘threaten’ontheotherisamatterofselectionalrestrictions. While beginnen ‘begin’cantake eithera vPorVP, versprechen ‘promise’and drohen ‘threaten’canonlytakea vP.Alternatively, onemayanalyze(43)ashavingamonoclausalstructure,inwhich beginnen ‘begin’occupiesa positionbetweenthepassive vandV,i.e.LAspundertheproposedanalysisofEAVs.Inthis analysis, beginnen ‘begin’istheonlyambiguousverbthatallowsforlongpassivebecauseitis theonlyonethatcanheadanLAspphrase.WhileIamnotawareofanyfurtherevidencethat wouldteaseapartthesetwopossibleanalysesoflongpassive,itseemssafetosaythattheaspect phrase analysis proposed in this study is a promising approach to accounting for the unique distributionof beginnen ‘begin’inGerman. 23 23 Wurmbrandinfactsuggeststhepossibilitythat beginnen ‘begin’isafunctionalhead(aspectualhead),althoughshe doesnotpursuethispossibility(Wurmbrand2001;fn.76).

26 AspectualVerbsandtheAspectPhraseHypothesis

4.2. Basque(ArregiandMolinaAzaola2004) Basqueofferssupportfortheproposedanalysisofaspectualverbs,foraverydifferentreason from the evidence presented so far. Arregi and MolinaAzaola (A&M) (2004) discuss two aspectual verbs, hasi ‘begin’ and amaitu ‘finish’, which show different agreement patterns. Whilebothaspectualverbsareanalyzedasrestructuringverbs,sincethematrixauxiliarycan agree with embedded arguments with both of these verbs 24 , only amaitu ‘finish’ allows the matrixauxiliarytoagreewithboththeembeddeddative and absolutiveargument(44a).Ascan be seen in (44b) and (44c), hasi ‘begin’ allows the matrix auxiliary to agree only with the embeddeddativeargument. (44) a. Berak [zuri babak egiten]amaitudautsuz heERG [youDAT beansABS doNF ]finished AGR AAGR DAGR E ABSDATERG ‘Hefinishedcookingthebeansforyou.’ (A&M2004;101,(1)) b. Bera [zuri babak egiten]hasi jatzu heABS [youDAT beansABS doNF ] began AGR DAGR A DATABS ‘Hebegancookingthebeansforyou.’(A&M2004;101,(2)) c.*Bera [zuri babak egiten]hasi jatzuz heABS [youDAT beansABS doNF ]began AGR AAGR D ABSDAT ‘Hebegancookingthebeansforyou.’(A&M2004;102,(3)) A&M attempt to associate the difference in agreement patterns and the fact that these two aspectualverbshavedifferentcaseontheirsubjects.Ascanbeseenabove, amaitu ‘finish’has an ergative subject (44a), while hasi ‘begin’ has an absolutive subject (44b). Descriptively, therefore,itappearsthatthematrixauxiliarycannotagreewithanembeddedargumentwitha particularcase,ifithasalreadyformedanagreementrelationwithamatrixargumentofthesame case. In (44c), the matrix auxiliary cannot form an agreement relation with the embedded absolutive argument, since it is already in an agreement relation with the matrix absolutive subject.Basedonthisobservation,A&Mhypothesizethatlocalityinagreementisrelativizedto case(A&M2004:108). Inordertoaccountforthedifferentagreementpatternsofthetwoaspectualverbs,A&M proposethatthesetwoaspectualverbsarefunctional headsthatassign absolutive case, which occupytwodifferentpositionsinaclause.Specifically,A&Marguethatwhile amaitu ‘finish’

24 Itisnotclearwhethertheseverbsarecontrolorraising.FromwhatIhavefoundintheliterature,however,these aspectual verbs appear tobe closer, intheir structural characteristics, to the verbs that are considered as controlin Basque,thantotheverbsthatareconsideredasraisinginBasque,asraisingverbsinBasquegenerallyrequirefinite complement(HualdeandOrtizdeUrbina2003:65356).

27 ShinFukuda

occupies the position immediately below vP, hasi ‘begin’ occupies the position immediately above vP(A&M2004:109,(17),(18)). (45) a. amaitu ‘finish’ b. hasi ‘begin’ TP TP V V vPTBeginPT V V v’ vPBegin V V FinishP v v’ V V VPfinishVP v 6 6 Theiranalysisaccountsforthetwoagreementpatternsasfollows.Oneoftheaspectualverbs, amaitu‘finish’providesabsolutivecasetotheembeddedverb’sdirectobject,sinceitiscloserto thedirectobjectthan v, theotherfunctionalheadwhichpotentiallyassignsabsolutivecase(46a). Incontrast,with hasi ‘begin’visclosertothedirectobject.Thus,itis vthatprovidesabsolutive casetothedirectobject,and hasi ‘begin’endsupprovidingitsabsolutivecasetothesubject.25 This accounts for the observation that only with hasi ‘begin’ is the subject marked with absolutive(46b). ABSERG (46) a. [[ vP Berak[ fin P [zuri babak Appl egiten]amaitu]v]T [[heERG [ [youDAT beansABS APPL doNF ] finish] v]T DAT ‘Hefinishedcookingthebeansforyou.’ ABSABS b. [beg P [vPBera[zuri babak Appl egiten] v]hasi]T [ [heABS [youDAT beansABS APPL doNF ]v]began]T DAT ‘Hebegancookingthebeansforyou.’ Assuming the relativized locality of agreement that they propose, this case assignment configurationaccountsfortheungrammaticalityofthelongdistanceabsolutiveagreementwith hasi ‘begin’in(44c). With amaitu ‘finish’,thematrixTcanagreewiththreearguments, the matrixergative,theembeddeddative,andtheembeddedabsolutive,asseenin(44a),sincethere arenotwoargumentsthatbearthesamecaseinthisconfiguration.Incontrast,with hasi ‘begin’, 25 Inbothconfigurations,A&Massumethatapplicativeheadisresponsibleforassigningdativecasetotheindirect object.

28 AspectualVerbsandtheAspectPhraseHypothesis

longdistanceagreementbetweenthematrixTandtheembeddedabsolutiveargumentisblocked by the matrix subject, which also bears absolutive case and is structurally higher than the embeddedabsolutiveargument. (47) TP V

BeginPT[ AGR A,AGR D] V vPbegin V Ext.arg v’ V VP v 6  IODOApplV WhatisinterestingaboutA&M’sanalysisofBasqueaspectualverbsforourpurposesisthatthey reachaverysimilarconclusiontotheproposedanalysisofEAVsaboutthesyntacticpositionsof the two aspectual verbs in Basque. There are two positions for aspectual verbs in Basque, immediatelybelowandabove vP,withoneoftheaspectualverbs, finish ,occupyingthelower positionandtheotheraspectualverb, begin ,thehigherposition.Importantly,theirconclusionis reachedbasedonaverydifferentsetofdataconcerningaverydifferentmechanismofgrammar thanwhatwehaveexamined,namely,agreement. 5. Conclusion Inthispaper,Iarguedagainstthecontrol/raisinganalysisofEnglishaspectualverbs(EAVs)and proposedanalternativeanalysis,inwhichEAVsprojecttheirownfunctionalprojections,aspect phrases,intwodifferentpositionsinaclause:between vandVP(LAsp)andabove vP(HAsp). IarguedthatthedifferenceinthepositionofEAVsmanifestsasthetwodifferentformsoftheir complements,gerundiveandinfinitive,andmotivatedtheproposedanalysisbasedonlanguage specificdataandcrosslinguisticdataconcerningsyntacticdistributionofaspectualverbs. WhileIhopetohaveshownthattheproposedanalysisofEAVsisapromisingapproachto analyzingaspectualverbs,theproposedanalysisalsoraisesanumberofquestionsthatthispaper didnotaddress.Forinstance,howarethedifferencesamongaspectualverbsacrosslanguages handled under the proposed analysis? While the proposed analysis captures the similarities amongaspectualverbsindifferentlanguages,theyalsohavedifferences.Anotherquestionthat theproposedanalysisraisesiswhethertheproposedanalysiscanbeextendedtootherpredicates withsimilarsyntactic behavior. Forinstance, modals and semimodal like predicates, suchas ‘want’and‘tend’,havealsobeenclaimedtoexhibittheambiguitybetweenathematicandnon thematic interpretation as well as monoclausal behavior in many languages. Could these predicates also be analyzed as functional heads within the verbal projection? If so, do they occupydifferentpositionsfromthepositionsforaspectualverbsordotheyappearinthesame syntacticpositions?Ihopethatthisstudyopensthedoortoinvestigationsintothesequestions and leads us to a better understanding of the syntax of the predicates that share similar characteristicswithaspectualverbs.

29 ShinFukuda

Totheextentitissuccessful,theproposedanalysisprovidesfurthersupportforthehypothesis that the functional projection inside the verbal projection encodes grammatical aspect, as originallyproposedinTravis(1991).Italsoenablesustoestablishaconnectionbetweenthe hypothesisthatthereisafunctionalprojectionwithintheverbalprojectionandthewellknown syntactic peculiarity of aspectual verbs and potentially other classes of semimodal like predicates.Finally,theapproachtothesyntaxofaspectualverbsadvancedinthisstudysuggests aparticularviewonnonfiniteclausalcomplementationingeneral.Constructionsthathavebeen analyzedasinstancesofnonfiniteclausalcomplementationmayinfactbeanalyzedassimple clauses,withpredicatesthat‘select’nonfiniteclausalcomplementbeinganalyzedasfunctional headsthatarepartsofthephrasestructureofsimpleclauses.Undersuchaview,whathasbeen analyzed as nonfinite clausal complementation does not involve embedding of clausal complementsinthetraditionalsenseafterall. References Agbayani,Brian&ChandraShekar.(toappear).RestructuringandclausalarchitectureinKannada.Toappearin CleverandRight:PapersinHonorofJosephEmonds . Aissen,Judith&DavidM.Perlmutter(1976).ClausereductioninSpanish. BLS 2.130. Aissen,Judith&DavidM.Perlmutter(1983).ClausereductioninSpanish.InDavidM.Perlmutter(ed.), Studiesin RelationalGrammar 1.360403.Chicago:TheUniversityofChicagoPress. Akmajian,Adrian,SusanM.Steele&ThomasWasow.(1979).ThecategoryAUXinUniversalGrammar.Linguistic Inquiry 10:164. Alexiadou,Artemis.(1997). AdverbPlacement:ACaseStudyofAntisymaticSyntax .Amsterdam:JohnBenjamins. Arregi,Karlos&GainkoMolinaAzaola.(2004).RestructuringinBasqueandthetheoryofagreement.WCCFL 23, 101114. Bobaljik,Jonathan.(1998).Floatingquantifiers:Handlewithcare[Stateofthe].GlotInternational 3.6:310. Bobaljik,Jonathan.(2003).FloatingQuantifiers:Handlewithcare.InLisaChengandRintSybesma,(eds.),The SecondGlotInternationalStateoftheArticleBook .MoutondeGruyter,107148. Bale,Alan.C.(2007).Quantifiersandverbphrases:Anexplorationofpropositionalcomplexity. aturalLanguages andLinguisticTheory 25:447483. Bolinger,Dwight.(1968).Entailmentandthemeaningofstructures. Glossa 2:119127. Bošković,Želiko.(1997). TheSyntaxofonfiniteComplementation:AnEconomyApproach .Cambridge,MA:MIT Press. Bresnan,Joan.(1972). TheoryofComplementationinEnglishSyntax. PhDdissertation.MIT. Brinton,Laurel,L.(1988). TheDevelopmentofEnglishAspectualSystems .NewYork:CambridgeUniversityPress. Cable,Seth.(2004). RestructuringinEnglish .Ms.,MIT. Chomsky,Noam.(1995). TheMinimalistProgram .Cambridge,MA:MITPress. Chung,Sandra.2004.RestructuringandverbinitialorderinChamorro.Syntax 7:199233. Cinque,Guglielmo.(1999). AdverbsandFunctionalHeads:ACrosslinguisticPerspective .Oxford:Oxford UniversityPress. Cinque,Guglielmo.(2003).Theinteractionofpassive,causative,and‘restructuring’inRomance.InChristinaTortora (ed.), TheSyntaxofItalianDialects .Oxford:OxfordUniversityPress. Collins,Chris.&HöskuldurThráisson.(1996).VPinternalstructureandobjectshiftinIcelandic. LinguisticInquiry 27,391444. Fischer,Susan&ByronA.Marshall.(1968). TheExaminationandAbandonmentoftheTheoryofBeginbyD.M. Perlmutteras ….Bloomington,IN:IndianaLinguisticCircle. Fitzpatrick,Justin.(2006). TheSyntacticandSemanticRootsofFloatingQuantification .PhDdissertation.MIT. Freed,AliceF.(1979). TheSemanticsofEnglishAspectualComplementation.London:Reidel. Fukuda,Shin.(2006).ThesyntaxofJapaneseaspectualverbs.Apaperpresentedatthe3rd WorkshoponAltaicin FormalLinguistics. Fukuda,Shin(2007).ProjectionsofaspectinJapanese.InYoichiMiyamotoandMasaoOchi(eds.), Formal ApproachestoJapaneseLinguistics:ProceedingsofFAJL4 .MITWorkingPapersinLinguistics55.3748. Givón,Talmy.(1973).Thetimeaxisphenomenon. Language 49:890925. Hallman,Peter.(2004).NPinterpretationandthestructureofpredicates. Language 80.707747.

30 AspectualVerbsandtheAspectPhraseHypothesis

Huarde,JoséIgnacio&JonOirtizdeUrbina.(2003). AGrammarofBasque .Berlin:MoutondeGruyter. Kageyama,Taro.(1993). Bumpotogokeisei [Grammarandwordformation].Tokyo:Hitsujishoboo. Kageyama,Taro.(1999).Wordformation.InNatsukoTsujimura(ed.), TheHandbookofJapaneseLinguistics .Oxford: Blackwell,297325. Koizumi,Masatoshi.(1994).ObjectagreementphrasesandthesplitVPhypothesis.InJonathanBobaljikandColin Philips(eds .),PapersonCaseandAgreementI .MITWorkingPapersinLinguistics18,99148. Koizumi,Masatoshi.(1995). PhraseStructureinMinimalistSyntax .PhDdissertation,MIT. Koizumi,Masatoshi.(1998).InvisibleagreeinJapanese. TheLinguisticReview 15:139. Kratzer,Angelika.(1994).Onexternalarguments.InElenaBenedicto&JefferyRunner(eds.), FunctionalProjections: UniversityofMassachusettsOccasionalPapers17,103130. Kratzer,Angelika.(1996).Severingtheexternalargumentfromtheverb.InJohanRooryck&LaurieZariing(eds.), PhraseStructureandthe Lexicon.Dordrecht:Kluwer. Landau,Idan.(2000). ElementsofControl:StructureandMeaninginInfinitivalConstructions .Dordrecht:Kluwer. Martin,Roger.(2001).NullcaseandthedistributionofPRO. LinguisticInquiry 32:141166. Matsumoto,Yo.(1996). ComplexPredicatesinJapanese:ASyntacticandSemanticStudyoftheotion‘word’ . TokyoandStanford:KuroshioandCSLI. May,Robert.(1985). LogicalForm:ItsStructureandDerivation .Cambridge,MA:MITPress. Newmeyer,FredericJ.(1975). EnglishAspectualVerbs .Paris:Mouton. Nishigauchi,Taisuke.(1993).Longdistancepassive.InNobukoHasegawa(ed.), JapaneseSyntaxinComparative Grammar.Tokyo:Kuroshio,79114. Perlmutter,DavidM.(1968). DeepandSurfaceStructureConstraintsinSyntax .Doctoraldissertation.MIT. Perlmutter,DavidM.(1970).Thetwoverbs begin .InRoderickA.Jacobs&PeterS.Rosenbaum(eds.), Readingsin EnglishTransformationalGrammar .Waltham,MA:Blaisdell,10719. Pesetsky,David.(1991).ZeroSyntaxII:AnEssayonInfinitives .Ms.,MIT. Potsdam,Eric.(inpress).AnalysingwordorderintheEnglishimperative.InWimvanderWurff(ed.),Imperative Clauses in Generative Grammar: Studies Offered to Frits Beukema. Amsterdam: John Benjamins, 111128. Pylkkänen,Liina&BrianMcElree.(2006).Thesyntaxsemanticinterface:onlinecompositionofsentence meaning. InMatthewTraxler&AnnGernsbacherMorton(eds.), HandbookofPsycholinguistics (2ndEd.).NewYork: Elsevier,537577. Rochette,Anne.(1999).Theselectionpropertiesofaspectualverbs.InKyleJohnson&IanRoberts(eds.),Beyond PrinciplesandParameters:EssaysinMemory ofOsvaldoJaeggli .Dordrecht:Kluwer. Ross,JohnR.(1972).Moreon begin . FoundationsofLanguage 8:574577. Shibatani,Masayoshi.(1973).Wheremorphologyandsyntaxclash:acaseinJapaneseaspectualverbs. GengoKenkyu 64:6596. Sportiche,Dominique.1990. Movement,CaseandAgreement .Ms.,UCLA. Sportiche,Dominique.1998.PartitionandAtomsofClauseStructure .London:Routledge. Stowell,Tim.(1982).Thetenseofinfinitives. LinguisticInquiry 13:561570. Thompson,Ellen.(2005). TimeinaturalLanguage:SyntacticInterfaceswithSemanticsandDiscourse .Berlin: MoutondeGruyter. Travis,LisaM.(1991). Derivedobject,inneraspect,andstructureofVP .PaperpresentedatNELS22. Wurmbrand,Susi.(2001). Infinitives:RestructuringandClauseStructure .Berlin:MoutondeGruyter. Wurmbrand,Susi.(2006).woll P:wheresyntaxandsemanticsmeet.PaperpresentedatNELS37. Wurmbrand,Susi.(2007).Infinitivesaretenseless.Toappearin UPennWorkingPapersinLinguistics 13.1. ShinFukuda DepartmentofLinguistics UniversityofCalifornia,SanDiego [email protected]

31